Clustered filesystem with data volume snapshot
View Patent ↗A computer system with read/write access to storage devices creates a snapshot of a data volume at a point in time while continuing to accept access requests to the mirrored data volume by copying before making changes to the base data volume. Multiple snapshots may be made of the same data volume at different points in time. Only data that is not stored in a previous snapshot volume or in the base data volume are stored in the most recent snapshot volume.
1. A method for recovering a viable quorum cluster of computers, the method comprising:
detecting a failed node in an initial viable quorum cluster by at least one node in the viable cluster when a member of the viable quorum cluster fails;
disabling message channels by the at least one node between the at least one node and the failed node after detecting the failed node;
disconnecting membership in the viable quorum cluster by the at least one node after disabling message channels after detecting the failed node;
blocking new nodes from joining the membership of the viable quorum cluster after detecting the failed node;
sending a withdrawal request from one node in the viable cluster to other nodes in the viable cluster when one node intends to deactivate from the viable quorum cluster in an orderly fashion;
electing a recovery leader;
initializing the recovery process;
initiating message interrupts;
initiating create locks;
preventing new accesses to objects associated with the recovery;
returning client tokens held;
purging states associated with a client;
thawing the message channels; and
releasing the create locks.
2. The method of claim 1 , further comprising:
establishing a new membership of the viable quorum cluster;
sending an acknowledgement to the node that requested withdrawal from membership; and
transitioning the node that requested withdrawal from membership to a shutdown state.
3. The method of claim 2 , further comprising upgrading software on the node previously in the shutdown state.
4. The method of claim 3 , further comprising:
entering the nascent state by the node previously in the shutdown state;
sending heartbeat signals from the node in the nascent state; and
the node in the nascent state joining the viable quorum cluster.
5. The method of claim 1 , further comprising:
flushing credentials of the failed node;
acknowledging that the credentials of the failed node have been flushed;
recovering the kernel object relocation engine;
recovering the volume manager;
recovering the filesystem; and
allowing new nodes to join the viable quorum cluster.
6. The method of claim 1 further comprising:
voting to elect a server host by remaining nodes;
selecting by the filesystem a next surviving node a possible server for the file system;
creating a manifest for each server on the failed node;
sending the manifest to the elected server host;
extracting information from the manifest;
sending a ready to commit instantiate callout message to the recovery leader; and
saving information relating to the extent of recovery.
7. The method of claim 6 , further comprising:
recovering the kernel object relocation engine in a first recovery pass;
recovering the volume manager in a second recovery pass; and
recovering the filesystem in a third recovery pass.